Re^10: Small Perl 6 discoveries II, Rats

[holli]

Yeah, well. That is to be expected as it is akin to assigning an int64 to an int32. What surprises me is that

perl6 -e "my FatRat $r = 0.1"

throws the same error.

[raiph]

First, Rats aren't FatRats and vice-versa. (If we need to discuss why, we'll need to start with Larry's comment that "we've intentionally steered clear of any sort of numeric tower"; and then dig into what p6curious meant when they followed that up with "its one of the things that has impressed me the most about perl6"; and then consider Rats and FatRats in particular.)

Second, P6 has a nice feature that directly addresses this. It has been implemented for routine parameters but not yet `my` declarations:

sub fails-if-passed-Rat    (FatRat   $r) { say WHAT $r } ;
sub works-if-passed-number (FatRat() $r) { say WHAT $r } ;
fails-if-passed-Rat    1.0 ; # typecheck failure
works-if-passed-number 1.0 ; # coerces 1.0 to a FatRat
my FatRat   $r = 1.0 ;       # typecheck failure
my FatRat() $r = 1.0 ;       # should say "not yet implemented"

[Anonymous Monk]

First, Rats aren't FatRats and vice-versa.

They're both Rational, but that's still not very useful.

> my Rational $x = 1.111111111111111111111;
1.11111111111111111604544
> $x *= 0.1;
Type check failed in assignment to $x; expected Rational but got Num (
+0.111111111111111e0)

The argument type-coercion is cute, but there doesn't seem to be any way to set the precision.

> sub foo(FatRat() $x) { return $x }; foo(pi).nude
(355 113)
> pi.FatRat(1e-8).nude
(103993 33102)

And I don't know how to interpret this message...

> my FatRat() $x = 0.1;
===SORRY!=== Error while compiling:
Coercion FatRat(Any) is insufficiently type-like to qualify a variable
------> my FatRat() $x⏏ = 0.1;
    expecting any of:
        constraint

[raiph]

In your first block of code you do a calculation whose resulting denominator is more than 64 bits. So the result is a Num.

Your second block shows BUG Num.FatRat coercion is inexact.

Your third block is an LTA error message. It should say something to the effect of "coercion type constraint on variable declaration not yet implemented".

Hth.